Run zero-shot image classification and cross-modal retrieval with CLIP.
npx skills add https://github.com/arsity/scholar-tools --skill clip-arsity
Please help me install this Agent Skill. Skill: clip Source: https://github.com/arsity/scholar-tools/tree/main/vendor/ai-research-skills/18-multimodal/clip Command: npx skills add https://github.com/arsity/scholar-tools --skill clip-arsity
OpenAI's CLIP provides a unified visual and textual embedding space enabling zero-shot image classification, image-text matching, and cross-modal retrieval without fine-tuning.
Install the CLIP package and run a sample zero-shot classification with your image.
High-intent search queries and answers about installing and using this skill.
Zero-shot image classification uses CLIP to map images and text prompts into a shared embedding space, scoring cosine similarity between image features and arbitrary label text to categorize images without fine-tuning.
Yes, CLIP enables cross-modal retrieval by encoding images and text queries into a unified visual-textual embedding space, allowing you to calculate image-text similarity for semantic image search workflows.
Running CLIP for image classification requires a PyTorch-based environment with the transformers, torch, and pillow dependencies installed to handle model inference, tensor operations, and image processing.
CLIP is applicable to content moderation workflows, providing zero-shot image-text matching to filter visual content against textual policy definitions without requiring task-specific fine-tuning.
CLIP calculates image-text similarity by running image and text encoders, applying feature normalization to the outputs, and computing cosine similarity scores across multiple available CLIP model variants.
Zero-shot CLIP provides rapid prototyping and baseline vision-language matching without fine-tuning, but may lack the specialized accuracy of trained models for domain-specific image classification or complex retrieval tasks.
ricableClassify images and score image-text similarity using CLIP embeddings.
eddieluengCompute image and text embeddings for zero-shot classification and retrieval.
hhhi21gCombine CLIP vision encoders with language models for image understanding tasks.
kaminocorpLoad OpenAI CLIP models to perform zero-shot image classification and image-text similarity.
